API Release Notes

Models launched:

• gemini-pro: New text model for a wide variety of tasks.
• gemini-pro-vision: New multimodal model.
• embedding-001: New embeddings model.
• aqa: Specially tuned model for answering questions using text passages for grounding.

API versions:

• v1: Stable API channel.
• v1beta: Beta channel with features under development.

API updates:

• GenerateContent: Single unified endpoint for chat and text.
• Streaming via StreamGenerateContent.
• Multimodal capability: Image support.
• New beta features: Function Calling, Semantic Retriever, Attributed Question Answering (AQA).
• Updated candidate count: Gemini models return only 1 candidate.
• New Safety Settings and SafetyRating categories.